Exact Fractional Inference via Re-Parametrization & Interpolation between Tree-Re-Weighted- and Belief Propagation- Algorithms

Inference efforts -- required to compute partition function, , of an Ising model over a graph of ``spins" -- are most likely exponential in . Efficient variational methods, such as Belief Propagation (BP) and Tree Re-Weighted (TRW) algorithms, compute approximately minimizing respective (BP- or TRW-) free energy. We generalize the variational scheme building a -fractional-homotopy, , where and correspond to TRW- and BP-approximations, respectively, and decreases with monotonically. Moreover, this fractional scheme guarantees that in the attractive (ferromagnetic) case , and there exists a unique (``exact") such that, . Generalizing the re-parametrization approach of \citep{wainwright_tree-based_2002} and the loop series approach of \citep{chertkov_loop_2006}, we show how to express as a product, , where the multiplicative correction, , is an expectation over a node-independent probability distribution built from node-wise fractional marginals. Our theoretical analysis is complemented by extensive experiments with models from Ising ensembles over planar and random graphs of medium- and large- sizes. The empirical study yields a number of interesting observations, such as (a) ability to estimate with fractional samples; (b) suppression of fluctuations with increase in for instances from a particular random Ising ensemble.
View on arXiv